A GMM/HMM model for reconstruction of missing speech spectral components for continuous speech recognition

This paper presents a method for reconstructing unreliable spectral components of speech signals using the statistical distributions of the clean components. Our goal is to model the temporal patterns in speech signal and take advantage of correlations between speech features in both time and frequency domain simultaneously. In this approach, a hidden Markov model (HMM) is first trained on clean speech data to model the temporal patterns which appear in the sequences of the spectral components. Using this model and according to the probabilities of occurring noisy spectral component at each states, a probability distributions for noisy components are estimated. Then, by applying maximum a posteriori (MAP) estimation on the mentioned distributions, the final estimations of the unreliable spectral components are obtained. The proposed method is compared to a common missing feature method which is based on the probabilistic clustering of the feature vectors and also to a state of the art method based on sparse reconstruction. The experimental results exhibits significant improvement in recognition accuracy over a noise polluted Persian corpus.     

What can possibly go wrong? Anticipatory work in space operations

This paper explores how different forms of anticipatory work contribute to reliability in high-risk space operations. It is based on ethnographic field work, participant observation and interviews supplemented with video recordings from a control room responsible for operating a microgravity greenhouse at the International Space Station (ISS). Drawing on examples from different stages of a biological experiment on the ISS, we demonstrate how engineers, researchers and technicians work to anticipate and proactively mitigate possible problems. Space research is expensive and risky. The experiments are planned over the course of many years by a globally distributed network of organizations. Owing to the inaccessibility of the ISS, every trivial detail that could possibly cause a problem is subject to scrutiny. We discuss what we label anticipatory work: practices constituted of an entanglement of cognitive, social and technical elements involved in anticipating and proactively mitigating everything that might go wrong. We show how the nature of anticipatory work changes between planning and the operational phases of an experiment. In the planning phase, operators inscribe their anticipation into technology and procedures. In the operational phase, we show how troubleshooting involves the ability to look ahead in the evolving temporal trajectory of the ISS operations and to juggle pre-planned fixes along these trajectories. A key objective of this paper is to illustrate how anticipation is shared between humans and different forms of technology. Moreover, it illustrates the importance of including considerations of temporality in safety and reliability research.     

Exploring and exploiting a historical corpus for Arabic

This paper presents a historical Arabic corpus named HAC. At this early embryonic stage of the project, we report about the design, the architecture and some of the experiments which we have conducted on HAC. The corpus, and accordingly the search results, will be represented using a primary XML exchange format. This will serve as an intermediate exchange tool within the project and will allow the user to process the results offline using some external tools. HAC is made up of Classical Arabic texts that cover 1600 years of language use; the Quranic text, Modern Standard Arabic texts, as well as a variety of monolingual Arabic dictionaries. The development of this historical corpus assists linguists and Arabic language learners to effectively explore, understand, and discover interesting knowledge hidden in millions of instances of language use. We used techniques from the field of natural language processing to process the data and a graph-based representation for the corpus. We provided researchers with an export facility to render further linguistic analysis possible.     

Cross level semantic similarity: an evaluation framework for universal measures of similarity

Semantic similarity has typically been measured across items of approximately similar sizes. As a result, similarity measures have largely ignored the fact that different types of linguistic item can potentially have similar or even identical meanings, and therefore are designed to compare only one type of linguistic item. Furthermore, nearly all current similarity benchmarks within NLP contain pairs of approximately the same size, such as word or sentence pairs, preventing the evaluation of methods that are capable of comparing different sized items. To address this, we introduce a new semantic evaluation called cross-level semantic similarity (CLSS), which measures the degree to which the meaning of a larger linguistic item, such as a paragraph, is captured by a smaller item, such as a sentence. Our pilot CLSS task was presented as part of SemEval-2014, which attracted 19 teams who submitted 38 systems. CLSS data contains a rich mixture of pairs, spanning from paragraphs to word senses to fully evaluate similarity measures that are capable of comparing items of any type. Furthermore, data sources were drawn from diverse corpora beyond just newswire, including domain-specific texts and social media. We describe the annotation process and its challenges, including a comparison with crowdsourcing, and identify the factors that make the dataset a rigorous assessment of a method’s quality. Furthermore, we examine in detail the systems participating in the SemEval task to identify the common factors associated with high performance and which aspects proved difficult to all systems. Our findings demonstrate that CLSS poses a significant challenge for similarity methods and provides clear directions for future work on universal similarity methods that can compare any pair of items.     

Safe Autonomous Agent Formation Operations Via Obstacle Collision Avoidance

In this paper, we consider the problem of flocking and shape‐orientation control of multi‐agent systems with inter‐agent and obstacle collision avoidance. We first consider the problem of forcing a set of autonomous agents to form a desired formation shape and orientation while avoiding inter‐agent collision and collision with convex obstacles, and following a trajectory known to only one of the agents, namely the leader of the formation. Then we build upon the solution given to this problem and solve the problem of guaranteeing obstacle collision avoidance by changing the size and the orientation of the formation. Changing the size and the orientation of the formation is helpful when the agents want to go through a narrow passage while the existing size or orientation of the formation does not allow this. We also propose collision avoidance algorithms that temporarily change the shape of the formation to avoid collision with stationary or moving nonconvex obstacles. Simulation results are presented to show the performance of the proposed control laws.     

Comparative evaluation of performance measures for human driving skills

In this paper, we evaluate the adequacy of several performance measures for the evaluation of driving skills between different drivers. This work was motivated by the need for a training system that captures the driving skills of an expert driver and transfers the skills to novice drivers using a haptic-enabled driving simulator. The performance measures examined include traditional task performance measures, e.g., the mean position error, and a stochastic distance between a pair of hidden Markov models (HMMs), each of which is trained for an individual driver. The emphasis of the latter is on the differences between the stochastic somatosensory processes of human driving skills. For the evaluation, we developed a driving simulator and carried out an experiment that collected the driving data of an expert driver whose data were used as a reference for comparison and of many other subjects. The performance measures were computed from the experimental data, and they were compared to each other. We also collected the subjective judgement scores of the driver’s skills made by a highly-experienced external evaluator, and these subjective scores were compared with the objective performance measures. Analysis results showed that the HMM-based distance metric had a moderately high correlation between the subjective scores and it was also consistent with the other task performance measures, indicating the adequacy of the HMM-based metric as an objective performance measure for driving skill learning. The findings of this work can contribute to developing a driving simulator for training with an objective assessment function of driving skills.     

A review of interoperability assessment models

Interoperability is the ability of systems to provide services to and accept services from other systems, and to use the services exchanged so as to operate together in a more effective manner. The fact that interoperability can be improved means that the metrics for measuring interoperability can be defined. For the purpose of measuring the interoperability between systems, an interoperability assessment model is required. This paper deals with the existing interoperability assessment models. A compara- tive analysis among these models is provided to evaluate the similarities and differences in their philosophy and implementation. The analysis yields a set of recommendations for any party that is open to the idea of creating or improving an interoperability assessment model.     

Sound source localization using compressive sensing-based feature extraction and spatial sparsity

In this paper, we propose a source localization algorithm based on a sparse Fast Fourier Transform (FFT)-based feature extraction method and spatial sparsity. We represent the sound source positions as a sparse vector by discretely segmenting the space with a circular grid. The location vector is related to microphone measurements through a linear equation, which can be estimated at each microphone. For this linear dimensionality reduction, we have utilized a Compressive Sensing (CS) and two-level FFT-based feature extraction method which combines two sets of audio signal features and covers both short-time and long-time properties of the signal. The proposed feature extraction method leads to a sparse representation of audio signals. As a result, a significant reduction in the dimensionality of the signals is achieved. In comparison to the state-of-the-art methods, the proposed method improves the accuracy while the complexity is reduced in some cases.